Data Synchronization in a Distributed Data Storage System

2. The method of claim 1, wherein the synchronization comprises identifying missing or changed data items and redistributing them among the nodes.

3. The method of claim 1, wherein the set of data items is synchronized in such a way that for all non-deleted data items, newer versions of the data items always replace older.

4. The method of claim 1, wherein the set of data items is synchronized in such a way that deleted data items always replace none-deleted, even if they are older.

5. The method of claim 1, wherein the information regarding missing or changed data items is received in the form of snapshots comprising information relating to the set of data items at a snapshot time Ts which are received from all other nodes in the cluster, said information identifying changes to the data items where each change comprises a time stamp, where only changes to the data items with a time stamp≤T0, where Ts>T0 and T0 is determined in such a way that it for the selected snapshot time Ts represents a corresponding time in all nodes, are included in the snapshot, although changes to the data items up to the time Ts have been received and stored in the nodes, and said node is determined not to be synchronized with the other nodes in the cluster if a corresponding snapshot generated locally in the node does not match all the received snapshots.

6. The method of claim 5, wherein T0 is calculated according to the equation T0=Ts−(Ts mod Td)−Td, where Td is a delay time set to be longer than the maximum transmission delay time between the nodes in the system.

7. The method of claim 5, wherein the synchronization state is determined by calculating a checksum based on said information relating to said data items, where the checksum is generated by merging generated and received information.

8. The method of claim 5, wherein the snapshots are generated only if the time stamps of at least one of the data items is older than a predefined snapshot updating time Tu.

9. The method of claim 5, wherein the snapshots are generated periodically, the frequency of generation being based on the frequency of changes to the data items.